This application is a revision which we believe is completely responsive to the recent review of our competitive renewal. Additional progress since the previous submission and Site Visit are emphasized. Rationale for a five (rather than four) year budget is emphasized. Key themes in the revision maintain program interactions among cell biologists, biochemists, molecular biologists and clinicians, that have extended key postulates in our investigations of the pathobiology of human colorectal cancer during the past five years. Broad experimental questions are still based on clinical phenomena: (l) What are the structure/function relationships in integrin laminin receptors and other adhesion molecules that may help to explain the unique biology of poorly differentiated or highly mucin producing colorectal carcinoma? Can these determinants of aggressive behavior predict bad outcome among the bore prevalent and heterogeneous moderate-to-well differentiated morphologies? (2) Can potential "founder genes" such as MSH2 and MLH1 be functionally defined in new model systems that help to explain their effects not only in HNPCC but also possibly "sporadic" colorectal carcinoma? (3) Can reconstitution models allow us to link many of the genetic changes found to be associated with gastrointestinal epithelial transformation to function and phenotype in premalignant or malignant mucosa? (4) Can information from our cDNA subtractive libraries be defined more elegantly by palindromic PCR driven differential DNA display in searching for significant mRNA differences between primary human colorectal carcinoma and adjacent epithelium and between primary and metastatic colorectal carcinoma? Finally, (5) Can markers studied during the last four years such as sucrase-isomaltase, DF3, Mac-2, CD44, and alpha6beta4 integrin be applied in archival tissues and in our prospective tissue bank to predict better the biologic aggressiveness of colorectal carcinoma?